JP4343208B2 - Relief activation method, relief activation device, relief activation program - Google Patents

Description

  The present invention relates to a rescue activation method, a rescue activation device, and a rescue activation program for performing a process and server failure restart (and recovery) process in a system in which a plurality of processes cooperate with each other on a plurality of servers to execute a plurality of tasks.

In the conventional server restart function, all servers are restarted as the next phase of the single server restart, and there is a concern about the influence on the entire system due to the influence of the interruption time. In addition, for process execution management, a method has been adopted in which a corresponding process is restarted when a failure is detected, or a process on the corresponding server is restarted on another server when a server fails. In these methods, the initialization process of the restarted process is executed, but the initialization process of the related process is not executed, and detailed settings cannot be made even if restarting to another server. There is concern that inconsistencies between processes will occur. As a result, the consistency of the entire system is lost, and the entire system is restarted to avoid an event, which may increase the interruption time.
In addition, in the prior art, a document that directly indicates the present invention has not been found and cannot be clearly indicated.

  As described above, in the conventional server restart method, restart of another server is forced to restart another server. A specific example is shown in FIG. When the user wants to restart specific servers SV-1 and SV-2, considering the operational relationship of other servers SV-3, SV-4, and SV-5 with respect to servers SV-1 and SV-2 Then, the entire system must be restarted and the servers SV-1 and SV-2 can be matched with the other servers SV-3, SV-4, and SV-5.

  In addition, due to a certain process failure, the consistency of the entire system is lost, and the entire system must be restarted to avoid the event. Specific examples thereof are shown in FIGS. First, in FIG. 16, a process (A) and a process (B) are functionally related processes, and in order to operate each, it is assumed that mutual operation is an essential condition. If one process is restarted due to a failure, a failure resulting from various contradictions occurs, and the other process must be restarted. The only way to achieve this is to restart the system.

  Next, in FIG. 17, a process (C) is a process functionally related to another commercial application, and the operation of the commercial application 10 is an indispensable condition for operating the system. If a failure occurs in the commercial application 10, execution management cannot be performed for the commercial application 10, so the system must be restarted.

  Next, in FIG. 18, process (A), process (B), process (C), and process (D) are functionally related processes, and in order to operate each other, mutual operation is essential. Suppose that it is a condition. When the restart of one server SV-1 occurs, the process (A) and the process (B) are restarted. Therefore, considering the occurrence of a failure caused by various contradictions, the process (C) on the server SV-2 side , It is also necessary to restart the process (D). In order to realize this, the server SV-2 can only be restarted.

  Next, regarding the recovery method of the startup process in the event of a server failure, detailed exclusive control setting cannot be performed, and the recovery startup of the corresponding process cannot be performed. This also necessitates restarting the entire system. A specific example is shown in FIG. In FIG. 19, it is assumed that the process (A) and the process (B) are in a relationship incapable of being activated on the same server in each operation. When the restart of the server SV-1 occurs, the rescue activation to the server SV-2 is impossible for the process (A) activated on the server SV-1 by the activation of the process (B). However, the activation to the server SV-3 is logically possible. However, since detailed exclusive setting for each process and server cannot be performed, the process (A) cannot be rescued and activated.

  In the conventional server restart method, any of these three events has a problem of prolonging the system operation interruption time.

  The present invention has been made in view of the above-described circumstances, and it is possible to prevent some failures from developing into resumption (interruption) of the entire system, and it is possible to reduce the interruption time of the entire system. It is an object of the present invention to provide a startup method, a relief startup device, and a relief startup program.

Further, in order to solve the above-described problems, the present invention relates to a process for rescue startup of a managed system in which a plurality of servers, a plurality of processes, or both are constructed in cooperation with each other. process excluding condition to disable the activation server excluding condition to disable the activation process of both server process group excluding condition to disable the activation of both the process group in the same server, and both the server group If a server group exclusion condition that disables activation is defined as exclusion condition information, and one of the plurality of servers or the plurality of processes fails and cannot be resumed, another condition is determined based on the exclusion condition information. From the servers that are still in operation, the process that was started on the failed server or the recovery of the failed process Start and characterized in that to identify the server as possible.

  The present invention is characterized in that, in the above-described invention, when there are a plurality of servers that can be rescue activated, a server with a small total number of activation processes is identified as a rescue activation destination.

In addition, in order to solve the above-described problems, the present invention provides a plurality of servers, a plurality of processes, or both processes on the same server in a rescue startup device of a managed system that is constructed in cooperation with each other. process excluding condition to disable the activation server excluding condition to disable the activation process of both server process group excluding condition to disable the activation of both the process group in the same server, and both the server group If the exclusion condition information storage means for storing server group exclusion conditions that cannot be activated as exclusion condition information and any of the plurality of servers or the plurality of processes fail and cannot be resumed, the exclusion condition information Based on the exclusion condition information of the storage means, it is started on the failed server from among other ongoing servers. It processes or relief boot failed process, characterized by comprising a server specifying means for specifying the server as possible.

Further, in order to solve the above-described problems, the present invention provides a process exclusion condition that disables both processes on the same server , a server exclusion condition that disables the processes of both servers , and the same server. process group excluding condition to disable the activation of both the process group at, and the server group excluding condition to disable the activation of both the server group, the steps of defining as an exclusive condition information, the server or the number multiple of several If any of these processes fails and cannot be restarted, based on the exclusion condition information, the process that was started on the failed server or the repair of the failed process is selected from other ongoing servers. And a step of causing a computer to execute a step of identifying a server that can be started.

  In the present invention, for each of a plurality of processes, another process that needs to be restarted simultaneously with the failed process is defined as process group information in order to ensure consistency with the failed process. When a process fails due to a failure, a process group to be restarted is specified based on the process group information. Therefore, it is possible to restart the server within the minimum necessary range, to localize the range of influence when a process failure occurs, to ensure consistency in group units, and to use exclusive control considering the relationship between processes By the relief, it is possible to prevent a part of the failure from developing into resumption (interruption) of the entire system, and to shorten the interruption time of the entire system.

As described above, according to the present invention, realize a server restart in need minimal range and the range of influence upon process failure is localized, it guaranteeing the consistency of a group unit, and, The process remedy by exclusive control that takes into account the relationship between processes has the advantage that some failures can be prevented from developing into resumption (interruption) of the entire system, and the interruption time of the entire system can be shortened. It is done.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
A. Configuration of Embodiment FIG. 1 is a block diagram showing a configuration of a server process management system according to an embodiment of the present invention. In FIG. 1, the server process management system includes servers SV-1, SV-2, and SV-3. Server SV-1 starts process A, process B (commercial application), process C (commercial application), and management process X. The server SV-2 starts process D, process E, and management process Y. The server SV-3 starts process F, process G, and management process Z.

Process A and process B constitute process group PG1.
Process C, process D, process E, and process F constitute a process group PG2. Note that a process group is a process in which activation management is considered as a group of processes that are operationally related. The application range of the process group is not closed within one server, but is effective between all servers constituting the system. Further, the function of the present invention is not limited to a self-made process, but is effective for a commercial application.

  The server process management system according to the present embodiment includes a condition file having the configuration shown in FIGS. FIG. 2 is a conceptual diagram showing the configuration of the startup process management file 20. FIG. 3 is a conceptual diagram showing the configuration of the system configuration management file 21. FIG. 4 is a conceptual diagram showing the configuration of the exclusive control management file 22.

In order to set the server / process group in the startup process management file 20 and the system configuration management file 21 shown in FIG. 2 and FIG. 3, each is registered in advance in the management file. Settings are made for servers / processes that require management in operation, and an item for setting a group number is provided in each record item.
The management process makes a grouping decision for each process according to these settings. There is no upper limit to the number of settings for one group. When “0” is set in the group number, it is determined that the grouping is not set for the process and the server.

In FIG. 4, in order to set exclusive control related to process rescue activation, registration is performed in advance in the exclusive control management file 22. Regarding the setting method, there are the following four patterns. 1. 1. Exclusive setting for each server group 2. Exclusive setting for each server. 3. Exclusive setting for each process group This is an exclusive setting for each process. An identifier indicating which setting is set is set, and items for setting a target name and a target name for setting exclusive control are provided on the identifier. The management process determines the presence or absence of exclusive control according to these settings. There is no upper limit to the number of settings.
In addition, regarding the exclusive setting target name / subject name, a plurality of items can be set with the same name.

  This will be described in more detail below. In this embodiment, regarding a process that is activated in the managed server (group), when any process fails, another process that needs to be restarted at the same time as the process to ensure consistency with the process. Are extracted before operation of the managed server (group), and the extracted processes are collectively defined as a process group. Note that when defining a process group, a process group having an inclusion relationship may be defined, but a process group that shares only some processes is not defined.

  Further, the startup process management file 20 including at least three pieces of information of the startup process, the process group to which the startup process belongs, and the server on which the startup process is running is similarly operated by the managed server (group). Create before.

  If process groups having an inclusion relationship are defined, a plurality of process groups to which the startup process belongs can be defined. For example, information on the inclusion relationship between process groups (for example, “process group α includes process group β and process group ν”) and the minimum number of startup processes including the startup process as a process group corresponding to each startup process Two types of information are defined. Alternatively, as information corresponding to each activation process, two types of information are defined: process group information with the smallest number of activation processes including the activation process and all process group information other than the above including the activation process.

The management process for managing the managed server refers to the restart range using the information described above so that the number of restart startup processes is minimized by referring to the startup process management file 20 when an arbitrary process fails. Is specified, and the process of “Restart group including related processes” is realized.
When one process group information includes the other process group information, if a process included in the other process group information fails due to a failure, only the process in the other process group information is restarted. When a process included only in the process group information fails due to a failure, all processes belonging to one process group information including the other process group information are restarted.

  In addition to the startup process, the management process refers to the startup process management file 20 so that the number of restart startup processes is minimized when any managed server itself restarts. The resumption range is specified using the information, and the “group resumption including related processes” process is realized.

  Further, in the present embodiment, similarly to the startup process group in the managed server (group), the managed server itself also fails when any managed server fails in accordance with the system operation conditions. Extract other managed servers that need to be restarted at the same time as the management target server to ensure consistency with the managed server (s) before operating the managed server (s). Define a server group. When defining a server group, a server group having an inclusion relationship may be specified, but a server group that shares only some managed servers is not defined.

  Further, the system configuration management file 21 including at least two pieces of information of the management target server and the server group to which the management target server belongs is similarly created before the management target server (group) is operated.

  When server groups having an inclusion relationship are defined, a plurality of server groups to which managed servers belong can be defined. For example, information on the inclusion relationship between server groups (for example, “server group α includes server group β and server group”), and a managed server that includes the managed server as a server group corresponding to each managed server Two types of information are defined: the smallest number of server group information. Alternatively, as information corresponding to each managed server, two types of information, that is, server group information with the minimum number of managed servers including the managed server and all other server group information including the managed server are included. Define

The management process for managing the managed server uses the information described above so that the number of resume managed servers can be minimized by referring to the system configuration management file 21 when any managed server fails. To specify the restart range and implement the “server group restart” process.
When one server group information includes the other server group information, if a server included in the other server group information fails due to a failure, only the server in the other server group information is restarted. When a server included only in server group information fails due to a failure, all servers belonging to one server group information including the other server group information are restarted.

  Also, in this embodiment, when any managed server or boot process cannot be restarted, for boot process rescue boot, any server group with respect to at least four patterns of server group, server, process group, and process The startup process of managed servers belonging to the server extracts the exclusion conditions that indicate the relationship between processes, servers, process groups, and server groups that cannot be started on the same server before operating the managed server (s) And created as an exclusive control management file 22.

  Exclusion conditions include server group exclusion conditions for identifying server groups that cannot all be started, and server exclusion conditions for identifying managed servers that cannot start all the startup processes of any managed server Process group exclusion conditions for identifying process groups that cannot be started on the same managed server as the start process belonging to any process group, processes that cannot be started on the same managed server as any process There is a process exclusion condition to identify

  The management process during operation succession refers to the exclusive control management file 22 when a managed server failure occurs, identifies a startup process that can be repaired by each managed server, and performs a recovery startup process of the startup process. Realize.

B. Operation of Embodiment Next, the operation of the server process management system according to the present embodiment will be described. First, FIG. 5 and FIG. 6 are conceptual diagrams for explaining the specific operation principle of the group restart method (when a group process restart occurs). 5 and 6 show an outline of processing of each server when the process D fails.

  First, an outline of the operation on the server SV-2 will be described with reference to FIG. The management process Y on the server SV-2 detects the failure of the process D and refers to the startup process management file 20 (Sa1). Since the process D belongs to the process group PG2, a restart request for the process group PG2 is made to the servers SV-1 and SV-3, which are servers on which processes defined in the process group PG2 are started (Sa2). Then, simultaneously with the restart request to the other server, the process E defined in the process group PG2 in the own server is restarted (Sa3).

  Next, an outline of the operation of the server SV-1 will be described with reference to FIG. When the management process X on the server SV-1 receives the restart request for the process group PG2 from the server SV-2, it extracts the process of the process group PG2 operating on its own server from the startup process management file 20 (Sb1). ), The process C belonging to the process group PG2 is restarted (Sb2). Similarly, in the server SV-3, the process corresponding to the process group PG2 is restarted.

  Next, an outline of processing in the server SV-2 when the server SV-1 is restarted will be described. Here, FIG. 7 and FIG. 8 are conceptual diagrams for explaining the specific operation principle of the group restart method (when server restart occurs).

  First, an outline of the operation of the server SV-1 will be described with reference to FIG. When the server SV-1 is restarted due to a process failure or the like, the server SV-1 is restarted to each server, and then the server SV-1 is restarted. At the time of restart, the startup process management file 20 is referred to (Sc1), and the processes A, B, and C to be started on the own server are started (Sc2).

  Next, the processing outline of the server SV-2 will be described with reference to FIG. After receiving the server restart notification from the server SV-1, the startup process management file 20 is referred to (Sd1), and the process group PG1 and the process group PG2 operating on the server SV-1 operate on the own server. Check if it is. Therefore, since the group PG2 is targeted, the processes D and E in the process group PG2 are restarted (Sd2). Similarly, in the server SV-3, processes corresponding to the process group PG1 and the process group PG2 are restarted.

  Next, FIG. 9 is a block diagram showing a configuration example of server types (APL server, WWW server, DB server, etc.) and server groups. In the figure, a server group is a group of servers that are operationalally related to each server that constitutes the system. The server group SG1 is composed of an APL server SV-1 and a WWW server SV-2. The server group SG2 includes an APL server SV-3, an APL server SV-4, and a naming server SV-5. It is also effective to set servers of different types (types) to the same group.

  Next, FIG. 10 and FIG. 11 are conceptual diagrams for explaining the specific operation principle of the server group restart method according to the present embodiment. First, an overview of the process of server SV-1 that has received a restart request for server group SG1 from a maintenance person or the like will be described with reference to FIG. The management process of the server SV-1 references the system configuration management file 21 after receiving the restart request for the server group SG1 (Se1). The server SV-1 also performs resumption notification of the server group SG1 to the other server SV-2 (Se2), and performs resumption because the server SV-1 itself is also the server group SG1 (Se3).

  Next, an outline of server SV-2 server group restart processing will be described with reference to FIG. The only difference between the server SV-1 and the server SV-1 is whether the request source is a maintenance person or another server. After receiving the restart request for the server group SG1, the system configuration management file is referred to (Sf1). The server SV-2 belonging to SG1 is also restarted (Sf2).

  Next, FIG. 12 is a block diagram showing the relationship between the exclusive control target and the exclusive control target of the server group / server / process group / process based on the exclusive control management file in the rescue activation method when introducing the exclusive process control. . FIGS. 13 and 14 are conceptual diagrams for explaining an outline of exclusive control when the server SV-1 fails. Note that the exclusive control target / target is a server, server group, or process in order to control the startup of processes that cannot be allowed to start on the same server due to various restrictions in configuring the system. The control object is provided in the range of.

  In the management process of the server SV-3, when a failure of the server SV-1 is detected, the exclusive control management file 22 and the system configuration management file 21 are referred to in order to check whether the process can be relieved in the own server, and the server group Get exclusive information of server, process group and process. For the server group SG1 (SRVgrp-1), information on the servers constituting the server group is acquired from the system configuration management file 21, and the process group is acquired from the startup process management file 20. Then, according to these pieces of information, it is determined whether the own server is an exclusion target (Sg1).

  In this case, the server SV-3 belonging to the server group SG2 (SRVgrp-2) is not a process relief target because the exclusive control for the server group SG1 (SRVgrp-1) is described in the exclusive control management file. (Sg2).

  In the case of the server SV-5 that is not group-defined, there is no exclusive definition for the failure of the server SV-1 of the server group SG1 (SRVgrp-1), but the exclusive definition for the process group PG1 (PRCgrp-1) is the server SV. Since it is in the process group PG2 (PRCgrp-2) on −5, it is not a process relief target.

  Further, in the management process of the server SV-6, when a failure of the server SV-1 is detected, the exclusive control management file 22 and the system configuration management file 21 are referred to check whether the process can be relieved in the own server, Acquires exclusive information of server groups, servers, process groups, and processes. Then, from the viewpoint of the server group and the server, it is confirmed whether the own server is exclusively set for the server SV-1 (Sh1). In this case, it is determined that the rescue start of the process A on the server SV-6 is possible. However, since there is another server that can be rescue activated, in this case, the server SV-2, the total number of processes is compared based on the internal management information, and the server SV-2 with the smallest total number of activation is determined as the rescue activation destination ( Sh2).

It is a block diagram which shows the structure of the server process management system by embodiment of this invention. It is a conceptual diagram which shows the structure of a starting process management file. It is a conceptual diagram which shows the structure of a system configuration management file. It is a conceptual diagram which shows the structure of an exclusive control management file. It is a conceptual diagram for demonstrating the specific operation | movement principle of the group restart method (at the time of group process restart occurrence). It is a conceptual diagram for demonstrating the specific operation | movement principle of the group restart method (at the time of group process restart occurrence). It is a conceptual diagram for demonstrating the specific principle of operation of the group resumption method (at the time of server resumption occurrence). It is a conceptual diagram for demonstrating the specific principle of operation of the group resumption method (at the time of server resumption occurrence). It is a block diagram which shows the structural example of a server classification (APL server, a WWW server, DB server, etc.) and a server group. It is a conceptual diagram for demonstrating the specific principle of operation of the server group restart method by this embodiment. It is a conceptual diagram for demonstrating the specific principle of operation of the server group restart method by this embodiment. FIG. 10 is a block diagram showing the relationship between exclusive control targets and exclusive control targets of server groups, servers, process groups, and processes based on an exclusive control management file in the rescue activation method when process exclusive control is introduced. It is a conceptual diagram for demonstrating the exclusive control outline | summary when the server SV-1 fails. It is a conceptual diagram for demonstrating the exclusive control outline | summary when the server SV-1 fails. It is a conceptual diagram for demonstrating the problem of the process resumption method by a prior art. It is a conceptual diagram for demonstrating the problem of the server restart method by a prior art. It is a conceptual diagram for demonstrating the problem of the server restart method by a prior art. It is a conceptual diagram for demonstrating the problem of the server restart method by a prior art. It is a conceptual diagram for demonstrating the problem of the process relief starting system by a prior art.

Explanation of symbols

  20 startup process management file (process group information, process group information storage means) 21 system configuration management file (server group information, server group information storage means) 22 exclusive control management file (exclusive condition information, exclusion condition information storage means) SV- 1 to SV-6 server PG1, PG2 process group SG1, SG2 server group AJ process X, Y, Z management process (restart process specifying means, restart server specifying means, server specifying means)

Claims (4)

In the rescue start method of a managed system in which multiple servers or multiple processes or both are built in cooperation,
Process exclusion condition that prohibits both processes from starting on the same server, server exclusion condition that prohibits startup of processes on both servers, and process group exclusion condition that prohibits startup of both process groups on the same server And the server group exclusion condition that disables the activation of both server groups as exclusion condition information,
If any of the plurality of servers or the plurality of processes fails and cannot be restarted, a process started on the failed server from among other ongoing servers based on the exclusion condition information Or a relief activation method characterized by identifying a server capable of relief activation of a failed process.   2. The rescue boot method according to claim 1, wherein when there are a plurality of servers that can be rescue booted, a server with a small total number of boot processes is identified as a rescue boot destination among the servers. In the rescue start-up device of the managed system that is built by linking multiple servers or multiple processes, or both,
Process exclusion condition that prohibits both processes from starting on the same server, server exclusion condition that prohibits startup of processes on both servers, and process group exclusion condition that prohibits startup of both process groups on the same server , And a server group exclusive condition that disables activation of both server groups , and exclusive condition information storage means for storing as exclusive condition information,
If any of the plurality of servers or the plurality of processes fails and cannot be restarted, a faulty server is selected from other ongoing servers based on the exclusion condition information stored in the exclusion condition information storage unit. And a server identification unit that identifies a server that can perform rescue activation of a process that has been activated or a failed process. Process exclusion condition that prohibits both processes from starting on the same server, server exclusion condition that prohibits startup of processes on both servers, and process group exclusion condition that prohibits startup of both process groups on the same server Defining a server group exclusion condition that prohibits activation of both server groups as exclusion condition information,
If any of the multiple servers or multiple processes can not be resumed at fault, based on the exclusion condition information, among other operations ongoing server had been activated in the failed server process Or a step of identifying a server capable of remedy activation of a failed process.

Images